English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Score distillation sampling (SDS) and its variants have greatly boosted thedevelopment of text-to-3D generation, but are vulnerable to geometry collapseand poor textures yet. To solve this issue, we first deeply analyze the SDS andfind that its distillation sampling process indeed corresponds to thetrajectory sampling of a stochastic differential equation (SDE): SDS samplesalong an SDE trajectory to yield a less noisy sample which then serves as aguidance to optimize a 3D model. However, the randomness in SDE sampling oftenleads to a diverse and unpredictable sample which is not always less noisy, andthus is not a consistently correct guidance, explaining the vulnerability ofSDS. Since for any SDE, there always exists an ordinary differential equation(ODE) whose trajectory sampling can deterministically and consistently convergeto the desired target point as the SDE, we propose a novel and effective"Consistent3D" method that explores the ODE deterministic sampling prior fortext-to-3D generation. Specifically, at each training iteration, given arendered image by a 3D model, we first estimate its desired 3D score functionby a pre-trained 2D diffusion model, and build an ODE for trajectory sampling.Next, we design a consistency distillation sampling loss which samples alongthe ODE trajectory to generate two adjacent samples and uses the less noisysample to guide another more noisy one for distilling the deterministic priorinto the 3D model. Experimental results show the efficacy of our Consistent3Din generating high-fidelity and diverse 3D objects and large-scale scenes, asshown in Fig. 1. The codes are available athttps://github.com/sail-sg/Consistent3D.

Consistent3D: Towards Consistent High-Fidelity Text-to-3D Generation with Deterministic Sampling Prior